A few days ago Google released information on their new search algorithm update that they have called BERT. The idea is to better understand natural language in order to make search results far more intuitive in their replies.
Google have said that BERT will impact 1 in 10 of all search queries and is the biggest change in search since they released RankBrain almost five years ago.
The following article takes information delivered by Barry Schwartz of Search Engine Land in order to best explain what is important for you and importantly, how this will affect you.
BERT started to roll out at the beginning of last week and will be fully live shortly. They have detailed that it will initially be aimed at English language queries however, it will be expanded to other languages in the future. They have also said that it will be globally used with ‘featured snippets’ in all languages.
BERT is Googles neural network-based technique for natural language processing (NLP) pre-training and stands for Bidirectional Encoder Representations from Transformers. In short, BERT will help computers understand language like humans.
Google has said that BERT will help better understand the context of words within a user’s search and therefore will enable them to match those queries with more relevant results. It will also be used for featured snippets.
An example that Google used was if the search query was “2019 brazil traveller to usa need a visa,” the word “to” and its relationship to the other words in this query are important for understanding the meaning. Previously, Google wouldn’t understand the importance of this connection and would return results about U.S. citizens traveling to Brazil. With BERT, Search is able to understand this nuance and know that the common word “to” actually matters and that they can provide a far more relevant result for this query.
With Featured Snippets, an example provided by Google highlighted a more relevant featured snippet being returned for the query “Parking on a hill with no curb”. In the past, a query like this would confuse Google’s system as they placed too much importance on the word “curb” and ignored the word “no”. This meant they did not effectively understand how critical that word was to appropriately responding to this query. Therefore, they would return results for parking on a hill with a curb.”
Googles first method for understanding queries using artificial intelligence was called RankBrain and was released in 2015. It looked at both queries and the content of web pages in Google’s index to better understand what the meaning of each word was. BERT does not replace RankBrain, it is an additional method for understanding these content and queries. It is therefore added to Google’s ranking system.
RankBrain can and will still be used for some queries. But when Google thinks a query can be better understood with the help of BERT, Google will use that method. In fact, a single query can use multiple methods for understanding any query.
Google has always said that there are many ways it understands what the language within a search query means and importantly, how it relates this to content delivered on the Internet. BERT is simply another signal in order to better understand this language.
One important question that web managers could ask is ‘how can I optimise for BERT?’ The simple answer here is you cant – specifically! What you should be doing is delivering content for users as this is where Google is now driving its algorithm. This is Googles best effort to understand a searchers query and to therefore match this to a relevant result.
The question could be posed ‘why do we care?’ Well, do not forget that Google themselves have said this change represents one of the biggest leaps forward in five years of search. They have also said that some 10% of all queries will been impacted by this update.
It is therefore recommended that you check to see if your search traffic changes sometime in the next week as this will highlight how much your site is impacted by this change.
If you would like Pure Digital to analyse this for you then simply ask and we will be happy to look deeper for you.